Frequency Shaping, including the RIAA curve and mastering for LP
Most live recordings suffer from some frequency response imbalances, either a boominess in the bass due to reflections in the hall, or a lack of high frequencies due to generational loss from copying the tape over and over. There can also be problems with feedback, instruments out of balance, or the singer is too loud (or not loud enough). All of these problems can be fixed－or at least made less severe－through some careful frequency shaping.
Even most commercial studio productions are not ideally balanced, due to mastering practices (which we’ll get to in a moment).
Audacity has five ways to change the frequency balance in an audio file:
• A basic Bass and Treble control (a single control for each, centered at 1000 Hz with a fairly shallow slope)
• Four EQs (multi-band equalizers):
1. Filter Curve EQ, which is a graph 20-20kHz on which you can draw any line
from +30dB to -30dB at any frequency
2. Graphic EQ, a 31 band graphic EQ ± 20dB
3. AUGraphicEQ with 10 bands, ± 20dB
4. AUGraphicEQ with 31 bands, ± 20dB
Each of these has their own “best use.”
The Bass and Treble control is very simple, useful only for when your source material is seriously out-of-whack.
The equalizers provide more detailed frequency shaping. The 31-band equalizers are useful for filtering out feedback, which is always a single well-defined frequency (and sometimes its octave). If you have a spot of feedback in a recording, you can highlight just that section, and do a spectrum analysis on it to determine which frequency is feeding back. By filtering for just that frequency, very sharply and very narrowly, you can essentially remove the feedback (although sometimes it takes two passes) without affecting any of the instruments around it.
The 10-band equalizers, representing all ten octaves of audible sound, give you broad control over the whole range. They are useful for rolling off the high end (to remove tape hiss) or boosting the low end (to overcome bass deficiency). Sometimes (often…) an S-curve is necessary, where the very low bass is boosted, the high-bass is attenuated, the mid-range is subtly boosted and the extreme high end is rolled off. Each recording requires creating a custom curve, depending on the recording equipment and venue and the skill of the recording engineer. Once you have created a curve that “sounds natural” for a track, you can save it under Presets and Settings – User Presets and use it for all the tracks in a concert. Be prepared to “tweak and preview” several times before finding the right balance. And don’t be afraid to treat different sections individually, if needed.
If there are instruments out of balance with the rest of the band, you can select just their frequencies for attenuation in the places where they’re playing. We’ll also talk more about this in the discussion of “Volume Shaping.”
The RIAA Curve and Dolby™
Vinyl record albums (LPs) are encoded with the RIAA curve (Recording Industry Association of America). This boosts the treble and decreases the bass, because the lower the frequency the wider the record groove has to swing. In order to get the maximum playing time per LP side, the low frequencies have to be limited to the cutting head. On playback, the RIAA curve is inverted in the amplifier’s phono preamp, boosting the bass again and cutting the highs, hopefully bringing the balance back to somewhere near a flat frequency response. This explains why turntables cannot be hooked up to an AUX IN. In addition to the turntable’s very weak signal output, you need a dedicated phono preamplifier to decode the RIAA equalization.
The RIAA curve is not perfect. In addition to boosting the bass signals on the record, it also boosts any turntable rumble present. It will amplify any low frequency vibrations, including footsteps, if your turntable is not acoustically-isolated from the floor and your speakers. Last it boosts what’s called “groove rumble” which is the sound of the needle being pulled across the record. Groove rumble is out-of-phase left-to-right, and in some records hovers right around 100 Hz. Filtering can ameliorate it.
Similarly, Dolby Noise Reduction is used on almost all cassettes. Both Dolby-B and Dolby-C boost the high frequency inputs to the recorder (+20dB for C, +10dB for B), which are then de-emphasized on playback. Since the noise－the tape hiss of the cassette tape－stays constant, the noise should be reduced by 10 or 20 decibels while the signal is restored (theoretically) to flat.
When working with pre-existing sources like this, we always want to start with the maximum signal available. It is a lot easier to CUT a recording with too much treble or bass than it is to try to BOOST the treble or bass out of a sea of noise. That is why cassettes should always be digitized with the Dolby off, and LPs should be digitized with the RIAA curve on.
Mastering for LP
At this point it’s probably worth mentioning the process of mastering for LP. In addition to the RIAA curve used for limiting the low frequencies on vinyl albums, there is also a universal practice of filtering out all frequencies below 50Hz, for the same reason. To compensate for this, almost all LPs have a boost of several dBs just above the cutoff frequency, at about 64Hz. This makes the album sound “fuller” and “fatter” on small bookshelf speakers, which themselves have diminished response below around 100Hz.
Therefore when digitizing an LP to make a CD-R of it, you might want to “unmaster” it: CUT the mid-bass (~64Hz) and BOOST the low bass (below 50Hz) to flatten out the response. Your ears will tell you how much to cut or boost without it becoming audible as “an effect.”